Abstract

Characteristics of over 15 000 tropical cyclone (TC) inner cores are evaluated coincidentally using 37- and 85-GHz passive microwave data to quantify the relative prevalence of cold clouds (i.e., deep convection and stratiform clouds) versus predominantly warm clouds (i.e., shallow cumuli and cumulus congestus). Results indicate greater presence of combined liquid and frozen hydrometeors associated with cold clouds within the atmospheric column for TCs undergoing subsequent rapid intensification (RI) or intensification. RI episodes compared to the full intensity change distribution exhibit approximately an order of magnitude increase for inner-core cold cloud frequency relative to warm cloud presence. Incorporation of an objective ring detection algorithm shows the robust presence of rings associated with hydrometeors for 85-GHz polarization corrected temperatures () and 37-GHz vertically polarized brightness temperatures () for differentiating RI with significance levels ≥99.99%, while 37-GHz false color rings of a combined cyan and pink appearance surrounding a region that is not cyan or pink lack statistical significance for discriminating RI against lesser intensification. Rings of depressed and enhanced tied to RI suggest the combined presence of liquid and frozen hydrometeors within the atmospheric column, indicative of cold clouds. The rings also exhibit preferences for those with collocated more widespread ice scattering signatures to be more commonly associated with RI and general intensification.

1. Introduction

Improved understanding of tropical cyclone (TC) intensity change remains a focus of the tropical meteorology community. DeMaria et al. (2014) note that for many time scales intensity change predictions robustly improved in recent years; however, gains were typically weaker and not robust at ≤48 h. Across time scales, events occurring within the tails of the TC intensity change distribution inherently have reduced predictability. While rapid weakening episodes not associated with landfall are becoming increasingly well understood because of processes associated with secondary eyewall development (e.g., Houze et al. 2007; Rozoff et al. 2008; Wang et al. 2013) or environmental influences (e.g., Wood and Ritchie. 2015), rapid intensification (RI) episodes continue to lack causal consensus. Hendricks et al. (2010) reported a lack of statistically significant differences between a number of synoptic variables between TCs undergoing RI and those intensifying at a lesser rate, implying other processes appear to be causal for RI.

Investigation of inner-core processes in TC intensification and RI has seen the role of clouds and precipitation commonly highlighted in prior works. Clouds and precipitation exist within the ascending branch of the secondary circulation, and are associated with intensification via the following processes: maintenance and strengthening of the warm core via diabatic and adiabatic warming (e.g., Schubert and Hack 1982; Vigh and Schubert 2009; Harnos and Nesbitt 2016), convergence driving vorticity aggregation and upscale growth (e.g., Hendricks et al. 2004; Nguyen et al. 2008; Wang 2014; Harnos and Nesbitt 2016), and moistening of the inner core for subsequent convective development (e.g., Bister and Emanuel 1997; Wang 2014; Harnos and Nesbitt 2016). Cloud modes encompassing predominantly liquid processes (i.e., shallow cumuli and cumulus congestus) and those containing both liquid and frozen hydrometeors (i.e., deep convection and stratiform clouds) occur in the tropics (Johnson et al. 1999), while limited quantitative study of these cloud populations has taken place within TCs. By discounting synoptic variability for RI initiation, Hendricks et al. (2010) support the potential importance of cloud and precipitation processes in RI. Kaplan et al. (2010) also note robust differences for RI episodes for infrared brightness temperature () areal coverage ≤−30°C and standard deviation, associated with inner-core clouds and precipitation. Within the convective spectrum, so-called hot towers or convective bursts that penetrate the stratosphere are a common focus for TC intensification (e.g., Heymsfield et al. 2001; Kelley et al. 2004; Hendricks et al. 2004; Rogers 2010; McFarquhar et al. 2012), while less vertically developed convective modes have been less commonly studied. Recently, however, Wang (2014) has shown the presence and importance of shallow cumuli and cumulus congestus in TC genesis through moistening of the inner core and through the spin up of the low-level circulation. Terwey and Rozoff (2014) also have supported the presence of multiple modes of convection within the TC inner core throughout the TC life cycle. Tao and Jiang (2015) used Tropical Rainfall Measuring Mission (TRMM) Precipitation Radar (PR) observations to argue for the importance of precipitation with a 20-dBZ height <6 km in RI while showing little connection between precipitation with a 20-dBZ height ≥10 km and RI. Harnos and Nesbitt (2016) showed that for two simulated RI episodes under low and high wind shear, hot towers contributed the majority of the diabatic heating, moisture convergence, and potential vorticity tendency with secondary contributions from stratiform precipitation and substantially smaller roles for less vertically developed convection.

Remote sensing studies are often the favored choice when investigating TC clouds and precipitation for their ability to span beyond limited case studies. Passive microwave (PM) sensors prove an optimal remote sensing tool because of their lengthy data records, substantial numbers of platforms providing revisit times currently around 3 h (Hou et al. 2014), and the ability to discern microphysical characteristics based upon the frequency analyzed. PM frequencies between 10 and 37 GHz are sensitive to emissions by increased liquid water path (e.g., Wu and Weinman 1984; Spencer et al. 1989) while ≥30-GHz reductions are associated with scattering due to greater ice water path (e.g., Wilheit 1986; Vivekanandan et al. 1991). PM studies of TC intensification often relate s with current TC intensity (e.g., Adler and Rodgers 1977; Alliss et al. 1992; Rodgers et al. 1994; Rodgers and Pierce 1995; Cecil and Zipser 1999; Hoshino and Nakazawa 2007). Some studies also note the potential for PM s to indicate subsequent intensity changes (e.g., Rodgers and Adler 1981; Rao and MacArthur 1994; Rodgers and Pierce 1995; Rao and McCoy 1997; Cecil and Zipser 1999), despite correlations with future static intensity being typically superior. These studies do not account for precipitation structural morphology however, instead using s averaged over broad regions or from only the most extreme pixels that can miss critical changes to TC structure and organization. Some early efforts to incorporate PM structures into intensity change are summarized by Hawkins et al. (2001). Lonfat et al. (2004) evaluated spatial asymmetries in TRMM Microwave Imager (TMI) rain rates for global TCs over 1998–2000, noting a preference for rainfall in the forward quadrants of the TC relative to its motion while tropical storms exhibited greater asymmetries than hurricanes. Jones et al. (2006) incorporated a PM component across the TC’s innermost 100 km into the Statistical Hurricane Intensity Prediction Scheme (SHIPS) using the mean, maximum, minimum, and standard deviation, with standard deviation found to be a skillful predictor for the east Pacific. Overall, Jones et al. (2006) note PM inclusion in SHIPS improved intensity prediction by 4%–8%, with the greatest improvement found for TCs undergoing an intensity change of ±10 kt (1 kt = 0.5144 m s−1) over 24–48 h.

Much of the early work explicitly relating PM morphology to TC intensity change was led by R. Edson. Cocks et al. (1999) and Edson and Lander (2002) related structural patterns in 37- and 85-GHz imagery to stages of TC lifetimes. Edson (2004a) continued this work, introducing structures related to TC genesis and the interpretation of intensity from microwave images. Edson (2004b) used PM imagery for the first time to suggest the potential for RI prediction based on convective organization within the TC inner core. This was then substantially expanded upon by Edson and Ventham (2008) in linking RI onset to the appearance of a convective ring surrounding the TC center at 37 or 85 GHz along with at least one convective burst present within the ring.

Recently, two studies have supported the presence of wavenumber-1 symmetry of precipitation first introduced by Edson and Ventham (2008). Harnos and Nesbitt (2011, hereafter HN11) used 17 000+ PM overpasses to generate 85-GHz composites, sensitive primarily to frozen hydrometeors, by intensity change. These RI composites were stratified further by wind shear magnitude with composites of pixels relative to the TC center evaluated for the frequency of occurrence of s corresponding to modest and intense convection. Composites revealed a low-shear mode that contained an elevated probability of deep convection of “moderate intensity” surrounding the core in RI storms (with the highest frequency of deep convection to the left of the shear vector), while a rarer high-shear mode was associated with asymmetric, more vigorous convection occurring most frequently to the left of the shear vector. However, the composites of HN11 do not imply a consistent ringlike feature existing during particular storms (e.g., Barnes and Barnes 2014). A contrasting perspective to HN11 is given by Kieper and Jiang (2012, hereafter KJ12), who used 84 North Atlantic TCs to highlight the presence within a 37-GHz false color product [see Eq. (3) and associated discussion] of a predominantly cyan-colored ring encircling the TC center at some point during RI. When used in conjunction with SHIPS RI index values providing environmental information, the manually identified cyan ring occurred at some point during RI for 75% of cases. In their conclusions, KJ12 assert that symmetric warm rain is represented by the cyan ring and is accordingly critical to the RI process. The subjective nature of the ring algorithm undermines the potential utility of the KJ12 results, much like the infrared-based Dvorak technique (Dvorak 1975) can yield varied estimates of TC intensity based upon the observer’s interpretation (Velden et al. 1998). When using the 37-GHz false color product to quantify ring presence, there is uncertainty regarding what color is present in addition to the spatial pattern represented. Further, while HN11 suggest the ring may be a precursor to RI, KJ12’s cyan ring occurs at any point during RI, introducing the question of whether these features are interrelated and if they have predictive potential or are merely by-products of RI. It is also unclear what physical processes yield cyan in the 37-GHz false color product. We seek to investigate characteristics of cloud populations in cyan and noncyan regions to better understand the underlying cloud populations and relevant physics. Tao and Jiang (2015) attempt to reach a similar goal using PR data; however, they rely solely on radar 20-dBZ height values to classify precipitation. However, this approach is unable to differentiate convective from stratiform clouds, which could plausibly appear in their category of “shallow convection-precipitation” (20-dBZ height <6 km) based on numerous prior studies that show stratiform precipitation having typical a 20-dBZ height of 5–8 km within TCs (e.g., Black et al. 1994, 1996, 2002; Hence and Houze 2011; Didlake and Houze 2013). Thus, we seek to clarify the roles of cloud and precipitation within the TC inner-core using passive microwave observations and process-based radiative transfer modeling.

This article seeks to do the following:

  • provide an overview of cloud and physical process influences on PM s,

  • investigate the relative roles of warm and cold (i.e., containing both liquid and frozen hydrometeors within the atmospheric column) clouds within the TC inner core relative to intensity change, and

  • objectively quantify morphological perspectives of ringlike features of clouds and precipitation surrounding the TC inner core in various fields and their relationships with subsequent intensity change in an attempt to reconcile HN11 and KJ12.

2. Data and methods

The first sensor used is the Special Sensor Microwave Imager (SSM/I; Hollinger et al. 1987), a conically scanning PM radiometer aboard the Defense Meteorological Satellite Program’s F8F15 platforms. These satellites utilize a polar, sun-synchronous orbit sampling the same locations at consistent local times, aside from orbital drift impacts, resulting in several-hour data coverage gaps. Typically, two SSM/Is operate coincidently, allowing for increased sampling of TCs relative to platforms with only a single operating platform. The SSM/I operates at horizontal and vertical polarizations of 19.35, 37.0, and 85.5 GHz in addition to employing a 22.235-GHz vertically polarized channel. The 37- and 85-GHz channels are used here, with native resolutions of 37 km × 28 km and 15 km × 13 km, respectively. It is noted that the 37-GHz channel of the SSM/I is relatively coarse and may be prone to nonhomogeneous beam-filling effects within the TC inner-core, where radial gradients between clouds and cloud-free areas are typically sharp. All global SSM/I TC overpasses from 1987 to 2008 are included for study.

The second sensor used is the TMI (Kummerow et al. 1998), which is also a conically scanning PM radiometer, making it a logical extension of the SSM/I record. TRMM possesses an atypical orbit with coverage focused between ±35°, further making it ideal for TC studies. The TMI operates in horizontal and vertical polarizations at 10.65, 19.35, 37.0, and 85.5 GHz, plus a 21.3-GHz vertically polarized channel, with the near-identical frequencies to the SSM/I inviting merged-sensor analyses. Again, 37- and 85-GHz channels are used with resolutions of 16 km × 9 km and 7 km × 5 km prior to an August 2001 boost in the satellite’s orbit that degraded these values by approximately 15%. TMI overpasses of all global TCs from 1997–2008 are evaluated. While other PM sensors exist operationally in addition to SSM/I and TMI, differences in scanning geometries and operating frequencies introduce arguably greater uncertainties than those found between the SSM/I and TMI, where differences are exclusively resolution based.

Both PM records are processed to conform to the Hurricane Satellite-Microwave (HURSAT-MW) database, similar to the HURSAT data record (Knapp and Kossin 2007). Each overpass is centered on the 6-hourly best track position and then bilinearly gridded to a resolution of 8 km. The influence of downscaling the native resolution sensor data to the common 8-km grid remains unclear, but impacts will be felt most strongly at 37 GHz from SSM/I as a result of it possessing the coarsest native resolution. Best track data from the International Best Track Archive for Climate Stewardship (IBTrACS; Knapp et al. 2010), version 02r01 (v02r01), is used for TC position and wind speed information at 6-h intervals. With IBTrACS v02r01, when multiple forecasting agencies provide concurrent best track data for a TC, these values are averaged together. While different TC advisory centers use varied time-averaging periods for wind speed measurements, IBTrACS first applies a conversion factor for 1- to 10-min winds of 0.88 prior to averaging. Some uncertainty may be introduced into the observations via this conversion for advisory centers that do not routinely utilize uniform averaging periods. The grids are recentered via an objective scene interpretation algorithm, described in the  appendix. For inclusion in this study, each overpass must include a minimum of 50% swath coverage across the innermost 1° of the TC. All TCs poleward of 35° latitude are excluded as potential extratropical transition cases. TCs must have winds at the overpass time ≥34 kt for consideration. Intensity change definitions are applied to each PM overpass based upon the wind speed change over the subsequent 24 h (). These intensity change definitions and sample sizes are outlined in Table 1, with the RI definition from Kaplan and DeMaria (2003); < 30 kt are classified as non-RI. Shear information is incorporated via the NASA Modern Era Retrospective Reanalysis (MERRA; Rienecker et al. 2011) by evaluating the differences in winds over 850–200 hPa within a 200–800-km annulus. When composite plan views are constructed, Southern Hemisphere TCs are mirrored across the shear axis to ensure consistency in cyclonic orientation.

Table 1.

Intensity change definitions utilized in this study and their sample sizes.

Intensity change definitions utilized in this study and their sample sizes.
Intensity change definitions utilized in this study and their sample sizes.

3. PM interpretation and controls

In addition to s, several commonly utilized products derived from PM data are incorporated. The first are polarization corrected temperatures (PCTs), acting to correct for assumed levels of background water vapor and varied land surface emissivities to explicitly approximate hydrometeor contributions (Grody and Weng 2008). The relationship for 85-GHz PCT, , comes from Spencer et al. (1989), where

 
formula

with H and V indicating horizontal and vertical polarization, respectively, and the subscript indicating frequency. The 37-GHz PCT, , is similarly described by Cecil et al. (2002):

 
formula

Also used is the Naval Research Laboratory’s 37-GHz false color product of KJ12, which was initially described by Lee et al. (2002). This product combines , , and to generate the amount of red, green, and blue, respectively, for each satellite pixel to be colored, where

 
formula

The ranges of individual s overlaid with values and false color representation can be seen in Fig. 1. Realistic false color values predominantly fall into three regimes: pink, cyan, and other (generally appearing green or orange). As noted by Lee et al. (2002), pink is associated with “deep convection” while cyan is linked with “low-level water clouds and rain” while other colors typically correspond to a lack of hydrometeors or nonphysical s. In an effort to objectively delineate between false color regimes, the authors have subjectively defined the approximate boundaries for each as outlined in black in Fig. 1. Figure 2 details the procedure for determining which 37-GHz false color is represented dependent upon the values of and . It should also be noted that the decision tree in Fig. 2 differs from the process utilized by KJ12 and similar unpublished work (H. Jiang 2015, personal communication), and likely from those of Tao and Jiang (2015), who do not define cyan or pink ranges.

Fig. 1.

The and values overlaid with (white contours) and 37-GHz false color [shaded; see Eq. (3)]. Black lines delineate approximated color regimes of cyan (top right), pink (top left), and other (bottom). Black dashed lines indicate the bounds of the NRL false color product.

Fig. 1.

The and values overlaid with (white contours) and 37-GHz false color [shaded; see Eq. (3)]. Black lines delineate approximated color regimes of cyan (top right), pink (top left), and other (bottom). Black dashed lines indicate the bounds of the NRL false color product.

Fig. 2.

Decision tree for determination of 37-GHz false color regimes.

Fig. 2.

Decision tree for determination of 37-GHz false color regimes.

A PM example is shown for Hurricane Katrina from August 2005 (Fig. 3). Here, Katrina is a category 5 hurricane with 150-kt winds, exhibiting well-developed structure to ease interpretation. First displayed is (Fig. 3a), with the TC associated with elevated values because of emission by liquid water within the atmospheric column. There is little variability across Katrina (≤15 K) despite the expectation of stratiform and convective regions coexisting, as a result of the saturation of 37-GHz linearly polarized s for rain rates ≥4 mm h−1 (e.g., Wilheit 1986; McGaughey et al. 1996). Also noteworthy in is the sharp contrast of s between land (top portion of the figure; 285 K) and open ocean outside the TC (240 K). Although not shown, exhibits qualitatively similar behavior to . Because of surface emissivity variability at 37 GHz (e.g., Grody and Weng 2008 and Fig. 3a), and the false color product are subsequently evaluated solely for oceanic regions to limit the potential land surface variability influences. As shown in Fig. 3b, masks the land–ocean contrast, while more differentiation is seen in Katrina’s structure, with the inner-core and spiral rainbands discernible. Variability of for the TC occurs across 230–275 K, which is substantially broader than . Note regions with the warmest (Fig. 3a) correspond to reduced (e.g., the rainband extending south of Katrina and the eyewall region). This is due to predominantly responding to the scattering of upwelling radiation by large ice hydrometeors (i.e., graupel and hail), whereas the polarized channels respond predominantly to emissions by liquid hydrometeors. The combination of these two effects for intense convection typically results in increased but decreased as a result of the atmospheric column containing liquid hydrometeors below graupel and hail. The 37-GHz false color product (Fig. 3c) makes land surfaces appear cyan, open ocean green, and Katrina’s oceanic cloud regions vary between cyan and pink. Pink regions of Katrina’s clouds (e.g., 27°N, 88°W) align well with values < 270 K (Fig. 3b), suggesting a substantial presence of large ice hydrometeors. Katrina’s cyan regions (e.g., 27.5°N, 90°W) correspond to elevated (Fig. 3a) in the absence of a strong scattering signal (Fig. 3b), implying the presence of liquid hydrometeors absent substantial large ice. Finally, (Fig. 3d) reveals scattering associated with either smaller cloud ice than detected at 37 GHz or beam-filling effects of the finer resolution at 85 GHz, with depressed s approaching 180 K where the ice water path from frozen precipitation-sized hydrometeors is particularly large. The greater depressions and stronger gradients in relative to are indicative of a greater presence of ice hydrometeors that scatter at 85 GHz but not 37 GHz (Vivekanandan et al. 1991). Both and (not shown) are qualitatively similar to , except masks the land–ocean contrast, as with .

Fig. 3.

TMI overpass of Hurricane Katrina at 2133 UTC 28 Aug 2005 with winds of 150 kt. Displayed products are (a) , (b) , (c) 37-GHz false color, and (d) .

Fig. 3.

TMI overpass of Hurricane Katrina at 2133 UTC 28 Aug 2005 with winds of 150 kt. Displayed products are (a) , (b) , (c) 37-GHz false color, and (d) .

From Fig. 3 it remains unclear where quantitative distinctions lie between warm and cold clouds, while that analysis also neglects potential nonmicrophysical influences on s. Such nonmicrophysical influences can include low-level wind influences on surface roughness and adjustments to sea surface emissivity (e.g., Stogryn 1972; Schluessel and Luthardt 1991; En-Bo and Yong 2005). The Microwave Radiative Transfer Model (MWRT; Liu 1998) is used to further elucidate influences from hydrometeors and other possibilities. Two experiments are conducted to address to what degree 1) distinctions can be made between warm and cold clouds using and , 2) nonprecipitating “warm” cloud and rain could produce cyan (as defined herein) in false color imagery, 3) surface roughness due to a disturbed sea state in developing tropical cyclones could yield cyan, and 4) may be misinterpreted because of scattering from warm rain, not ice, particles. The moist tropical atmospheric sounding from Dunion (2011) is used, interpolated to mandatory levels at 50 hPa. Wind speed (sea surface temperature) is varied in each experiment from 0 to 50 m s−1 at 1 m s−1 intervals (299.15–305.15 K at 1-K intervals). In experiment 1, cloud liquid water [parameterized as in Deirmendjian (1964)] is specified uniformly at T < 273 K from 0 to 2 g kg−1 at 0.1 g kg−1 intervals. In experiment 2, rain rates [parameterized as in Rutledge and Hobbs (1983)] are varied between 0 and 20 mm h−1 at 1 mm h−1 intervals. The results from experiments 1 and 2 are shown in Figs. 4a–d and 4e–g, respectively.

Fig. 4.

(a) Simulated values (shaded) for experiment 1 as a function of surface wind speed (m s−1) and cloud liquid water (g kg−1). (b)–(d) Simulation results in the space of 37-GHz false color imagery, with green, cyan, and pink regions delimited by the black lines as in Fig. 1. Simulation points are colored by cloud liquid water in (b) (g kg−1), surface wind speed in (c) (m s−1), and sea surface temperature in (d) (K). (e)–(h) As in (a)–(d), but for experiment 2.

Fig. 4.

(a) Simulated values (shaded) for experiment 1 as a function of surface wind speed (m s−1) and cloud liquid water (g kg−1). (b)–(d) Simulation results in the space of 37-GHz false color imagery, with green, cyan, and pink regions delimited by the black lines as in Fig. 1. Simulation points are colored by cloud liquid water in (b) (g kg−1), surface wind speed in (c) (m s−1), and sea surface temperature in (d) (K). (e)–(h) As in (a)–(d), but for experiment 2.

Figures 4a and 4c display simulated as a function of wind speed for each experiment. Here, it is shown that at low cloud and rain liquid water path, the sea state is a strong control on . As the cloud and rainwater path increase above 0.2 g kg−1 and 2 mm h−1, respectively, the surface influence on is masked. Cloud liquid water and rain rate asymptote to near 273 and 255 K, respectively, with the latter due to scattering by raindrops at 85 GHz. This experiment shows that it is unlikely that < 250 K caused by shallow convection are widespread at satellite sensor resolution but, instead, are more likely caused by ice particles in vertically developed convective or stratiform precipitation. Thus, we subsequently distinguish ≤ 250 K in conjunction with ≥ 255 K (as per Fig. 3a) as representing cold clouds that include both liquid and frozen hydrometeors (e.g., deep convection, stratiform clouds). Warm clouds (i.e., shallow cumuli and cumulus congestus with minimal frozen hydrometeor presence) are defined as pixels having ≥ 255 K and ≥ 260 K. Ambiguous clouds are defined for between 250 and 260 K, where the potential frozen hydrometeor presence is less clear following the results here and in Fig. 14 in Spencer et al. (1989). We treat < 255 K as clear air, or as a relative absence of hydrometeors, following Figs. 3 and 4.

Figures 4b–d and 4f–h place the radiative transfer results within the framework of simulated , , and the 37-GHz false color products (shaded in the background of each image), with the regions of cyan and pink delimited as in Fig. 1. Each point from the simulation results is colored according to the parameter indicated in the color bar for each panel. In both experiments, increasing the cloud and rainwater path produces values that go from the green to the cyan region in 37-GHz false color imagery (Figs. 4b,f). At rain rates >10 mm h−1, weak 37-GHz scattering is seen, approaching pink in the false color imagery (Fig. 4f). However, results from varying the surface wind speed, while examining the same points in the hydrometeor values in the panels to their immediate left, indicate that increasing the surface wind above values corresponding to tropical storm to weak hurricane force can also provide for cyan at low cloud and rainwater paths. Sea surface temperature variations cause little change in the simulations (Figs. 4d,h). The results emphasize that when using a state-of-the-art radiative transfer model, it is ambiguous whether cyan in 37-GHz false color imagery (and thus the cyan ring signature of KJ12) is caused by cloud water in nonprecipitating clouds, rain, a disturbed sea surface, or some combination thereof.

4. Inner-core brightness temperature distributions

Before incorporating morphological aspects, a first approximation of the joint distributions of multifrequency s is sought. Cross-frequency analysis is used for collocated and to quantify the relative presence of liquid and frozen hydrometeors, with a bivariate probability distribution function (PDF) displayed in Fig. 5 at the individual pixel (8 km) scale within the innermost 1°. A logarithmic scale is utilized to better isolate extreme values within the distribution. While such analysis assumes vertical continuity, the 37- (85-) GHz weighting function peaks between 4–6 (8–10) km (Gasiewski 1993) and the typical vertical slope of eyewalls, assumed representative of the TC inner core, is 2–3 km between 5- and 9-km altitude (Hazelton and Hart 2013). Given that this is below the 8-km dataset resolution, expected influences are weak; however, it remains a caveat given the spread in the slope values of Hazelton and Hart (2013). The distribution in Fig. 5 is bimodal, with one peak near 0.23% (10−0.63%) occurring for ≥ 275 K and of 230–250 K, while a secondary peak exists between of 230–260 K and of 260–265 K. This first peak (constant , variable ) is associated with a relative absence of hydrometeors because of the cool and warm values. The second peak (constant , variable ) is associated with cold clouds (Figs. 3 and 4). This peak also displays the 37-GHz saturation effects noted by Wilheit (1986) and McGaughey et al. (1996). Broadening of the distribution (encompassed by orange and red values between these two PDF peaks) is associated with nonhomogeneous beam filling, varied background SSTs, and variations in frozen hydrometeor scattering character. It is also apparent from this figure that substantial ice scattering at can exist despite limited scattering at from large cloud ice, as seen in values ≤200 K occurring at near 260 K (bins in this vicinity each accounting for around 0.01% of the dataset). Pronounced scattering manifests rarely, typically for ≤ 125 K.

Fig. 5.

Two-dimensional PDFs of individual pixel occurrences of combined and across the innermost degree of all TCs. Bins are 1 K. Sample size is 8 504 220 pixels. Note color scale is logarithmic. Black lines denote boundaries between approximated cloud regimes (see text for more details).

Fig. 5.

Two-dimensional PDFs of individual pixel occurrences of combined and across the innermost degree of all TCs. Bins are 1 K. Sample size is 8 504 220 pixels. Note color scale is logarithmic. Black lines denote boundaries between approximated cloud regimes (see text for more details).

The cloud-type definitions introduced in section 3 are used to quantify contributions to the TC inner core in Fig. 5 (black lines). Using these approximations, the typical TC inner core is 24% cold clouds, 23% warm clouds, 9% ambiguous clouds, and 44% clear air. These values are within 2% of those from Tao and Jiang (2015) if the ambiguous and cold clouds in this study were approximately represented by the very deep, moderately deep, and moderate precipitation in Tao and Jiang (2015) and warm clouds here are assumed to be shallow precipitation from Tao and Jiang (2015). Slight discrepancies in these quantities may exist because of the PM geometry viewing off nadir. Reproducing Fig. 5 for solely SSM/I or TMI (not shown) reveals minor differences between the platforms, supporting the idea of sensor resolution differences not meaningfully impacting results within the framework.

Figure 5 can be subdivided via the intensity change definitions of Table 1, as seen in Fig. 6. Relative to the other panels in Fig. 6, the RI PDF (Fig. 6a) shows peak values near 0.20% (10−0.69%) of the distribution in the range of 225 ≤ ≤ 260 K and of 260–265 K. Here, scattering by frozen precipitation-sized hydrometeors is sufficient to reduce but there are minimal large ice hydrometeors (i.e., graupel and hail) or insufficient beam filling to depress . The maxima noted in the PDF for all intensity changes (Fig. 5) with values at ≥ 275 K, associated with an absence of hydrometeors, is absent in the RI distribution with the PDF shifting toward an increased hydrometeor presence. Using the same partitioning as for Fig. 5, it is found that for RI 39% of pixels are cold clouds, 25% warm clouds, 15% ambiguous clouds, and 22% clear air (note that the values do not sum to 100% because of rounding). While there is a 2% increase in the warm cloud occurrence for RI cases versus the full distribution, the comparable increase in cold clouds is over 7 times larger (15%). This result supports the presence and role of cold clouds for subsequent RI, in line with HN11. These values are disparate from those of Tao and Jiang (2015), who showed 29% coverage within the TC inner core for RI cases from shallow precipitation (7% increase relative to all cases) while moderate, moderately deep, and very deep precipitation accounted for 41% of the area (11% increase relative to all cases). The disparity between the relative increases here and those of Tao and Jiang (2015) is likely linked to the inability to discriminate stratiform clouds from shallow convection, yet their shifts are in line with the results here showing a more substantial increase for vertically developed clouds relative to shallow clouds tied to RI. Figures 6b–d show differences of the distributions for other intensity changes relative to RI in order to better isolate differences by intensity change. A similar quantification of cloud nature for the IN PDF (Fig. 6b) reveals 33% cold clouds, 23% warm clouds, 12% ambiguous clouds, and 32% clear air, again revealing the presence of increased frozen hydrometeor presence associated with TC intensification relative to the full distribution, yet less than for RI. These values are similar to those of Tao and Jiang (2015) if their very deep, moderately deep, and moderate precipitation results are assumed equivalent to cold clouds in this study. For SS and WE TCs (Figs. 6c,d) the PDFs shift toward lacking hydrometeors, while the peak associated with a joint presence of frozen and liquid hydrometeors seen in the full dataset, RI, and IN PDFs is muted. The warm cloud presence here is actually the largest for WE TCs (25%), which is at odds with the results of shallow convection from Table 4 in Tao and Jiang (2015), with the likely source again being their omission of differentiation between stratiform (containing ice) and shallow convective clouds (containing primarily liquid hydrometeors). It is also interesting to note that values <75 K do not occur in RI cases; thus, no link is apparent between extreme inner-core ice scattering and RI, as suggested by DeMaria et al. (2012).

Fig. 6.

As in Fig. 5, but for the innermost degree of (a) RI, (b) IN, (c) SS, and (d) WE TCs only. In (b)–(d) differences are shown with respect to RI. Sample sizes are 341 132, 1 702 447, 4 268 348, and 2 372 713 pixels, respectively.

Fig. 6.

As in Fig. 5, but for the innermost degree of (a) RI, (b) IN, (c) SS, and (d) WE TCs only. In (b)–(d) differences are shown with respect to RI. Sample sizes are 341 132, 1 702 447, 4 268 348, and 2 372 713 pixels, respectively.

Figures 5 and 6 call into question the claim of HN11 that minimal signal is found in the 37-GHz composites, as there may be elevated values associated with the reduced seen within the TC inner core. To that end, shear-relative composites are developed in Fig. 7, as in Fig. 1 of HN11. Instead of developing four composites for each intensity change classification as in HN11, white dots instead indicate that the corresponding pixel’s RI mean is not statistically significant at ≥99.9% relative to the corresponding IN mean using a two-sided t test. For (Fig. 7a), an annular region where differences are robust exists between approximately the 25- and 80-km radii, while the RI and IN composites (not shown) are qualitatively similar to HN11’s Fig. 1. At (Fig. 7b), the differences between RI and IN are robust over nearly the entire innermost 1°, with a region of elevated near 260 K at a radius of 50 km. Interestingly, differences are also robust between the RI and IN classifications near the TC center, while the RI and IN composites (not shown) support RI cases having warmer here than IN cases, which is suggestive of increased columnar liquid water. Warm near the center may also be an artifact of including high-shear RI episodes with strong convective asymmetries because of the influences of wind shear on the convective organization following HN11, Molinari et al. (2013), and Zhang et al. (2013).

Fig. 7.

Pixels of shear-relative RI composites of (a) and (b) . Composite s that are not statistically significant at or above 99.9% relative to the corresponding IN composites using a two-sided Student’s t test are indicated with a white dot. The shear vector is oriented toward the top of the page. Southern Hemisphere TCs are mirrored across the shear axis to account for cyclonic rotation differences. Range rings are ½°.

Fig. 7.

Pixels of shear-relative RI composites of (a) and (b) . Composite s that are not statistically significant at or above 99.9% relative to the corresponding IN composites using a two-sided Student’s t test are indicated with a white dot. The shear vector is oriented toward the top of the page. Southern Hemisphere TCs are mirrored across the shear axis to account for cyclonic rotation differences. Range rings are ½°.

To investigate the potential for high-shear cases to contaminate Fig. 7, shear-relative frequency composites are developed following HN11 at 6-h intervals during the 24 h preceding (negative times) and following (positive times) RI for ≤ 250 K, ≥ 255 K, and cyan presence when shear is <10 kt (Fig. 8). Frequencies are smallest for ≤ 250 K; however, by RI onset, there appears to be an annular feature surrounding the TC center at ≥50%–55% near 55-km radius. Near this radius following RI onset, the frequencies of ≤ 250 K are comparable to the mean eyewall completeness value of 64% in the aircraft radar observations of Barnes and Barnes (2014). These results qualitatively mirror those of HN11, supporting the usage of the recentering algorithm and the updated shear source having minimal impact on their findings. The ≥ 255-K frequencies are greater than ≤ 250 K, associated with an increased presence of liquid hydrometeors relative to frozen precipitation-sized hydrometeors. Higher frequencies of ≥ 255 K, relative to , encircling the TC center appear for all times relative to RI, increasing from 60%–70% to near 100% over the 48 h evaluated here. Cyan pixels occur at frequencies 5%–10% less than ≥ 255 K at comparable times and locations; however, an annular structure is lacking until +12 h. Before +12 h cyan frequencies tend to peak monotonically near the TC center, rather than having reduced frequencies near the TC center, as in and . Differences between the inner-core frequency contrasts at the TC center at ≥ 255 K support the cyan perspective here not being resolution driven. Further stratifying results into SSM/I- and TMI-only frequency perspectives (not shown) shows frequency differences of 5%–10% across sensors, while the spatial patterns are qualitatively consistent, reinforcing the lack of resolution impacts on Fig. 8. Variable eye widths may also impact Fig. 8, but the lack of an eye for all cases makes such influences difficult to discern.

Fig. 8.

Shear-relative frequency composites of 250 K (first column), 255 K (second column), and cyan false color presence (third column) at 6-h intervals over the 24 h immediately preceding and following RI onset for TCs undergoing RI with shear <10 kt. Sample sizes are indicated to the left of each set of columns. Shear vector is oriented toward the top of the figure (left of page). Southern Hemisphere TCs are mirrored across the shear axis to account for cyclonic rotation differences. Range rings are ½°.

Fig. 8.

Shear-relative frequency composites of 250 K (first column), 255 K (second column), and cyan false color presence (third column) at 6-h intervals over the 24 h immediately preceding and following RI onset for TCs undergoing RI with shear <10 kt. Sample sizes are indicated to the left of each set of columns. Shear vector is oriented toward the top of the figure (left of page). Southern Hemisphere TCs are mirrored across the shear axis to account for cyclonic rotation differences. Range rings are ½°.

Finally, high frequencies in Fig. 8 near the TC center at +24 h are noted for ≥ 255 K of 60%–65% and cyan pixels of 50%–55%. All TCs in the study must have a minimum wind speed at +24 h of 64 kt (minimum intensity of ≥34 kt initially and ≥30 kt gained during RI), the threshold for a category 1 hurricane, and slightly larger than the median value for initial eye development within the North Atlantic of 58 kt from Vigh et al. (2012). Some high frequencies within the centers of Fig. 8 may be due to potentially misplaced centers; however, the recentering algorithm makes this less likely, especially considering the relatively strong intensity and assumed accompanying greater structural organization. At +24 h, the peak frequencies near a ½° radius for and cyan pixels are 90%–95% and 70%–75%, respectively, yielding commensurate radial frequency gradients of near 30% and 20% relative to the TC center and eyewall region, less than the differences of 35%–40% differences seen for ≤ 250 K. The reduced radial frequency gradients, particularly for cyan frequencies, suggest the potential inability to discern any meaningful structural differences from near the circulation center and adjacent regions. A similar lack of radial frequency gradients is apparent in the combined cyan and pink frequency composite in Fig. 18 of Tao and Jiang (2015). Instead, it appears that the cyan coverage slowly grows throughout the TC inner core as RI progresses, albeit not to the extent that ≤ 250 K frequencies do near a radius of a ½° (55 km). Wind influences on 37-GHz s may be a source of this increase in and the cyan frequencies throughout RI (Figs. 4c,g).

For further delineation of the cloud populations across the innermost degree of TCs, Fig. 9 shows a quantile–quantile plot for each intensity change classification relative to all intensity changes for values from pixels of ≥ 255 K, ≥ 260 K, and cyan false color regions. Apparent in all panels in Fig. 9 is that the shapes of the distributions do not vary substantially and are largely linear, indicating limited differences in the distribution variances relative to intensity change. Of note is the shift to the points to the right as the intensity change definitions increase (i.e., WE farthest left and RI farthest right). This is indicative of the shift in the distribution mean values across intensity changes, with greater intensification linked to colder values associated with either ice scattering or nonhomogeneous beam-filling effects. A Kolmogorov–Smirnov test was performed on the individual intensity change distributions to determine whether the distributions can be represented by a common normal distribution. For RI versus IN distributions and RI versus non-RI distributions and distribution differences are statistically significant at ≥99% in all three 37-GHz products shown in Fig.9.

Fig. 9.

Quantile–quantile plots of (a) for ≥ 255 K, (b) ≥ 260 K, and (c) cyan regions across the innermost 1° of TCs by intensity change relative to the full intensity change distribution. Values are plotted at 2% intervals from 1% through 99% (closed circles), except for 5%, 15%, and so on, shown as open boxes.

Fig. 9.

Quantile–quantile plots of (a) for ≥ 255 K, (b) ≥ 260 K, and (c) cyan regions across the innermost 1° of TCs by intensity change relative to the full intensity change distribution. Values are plotted at 2% intervals from 1% through 99% (closed circles), except for 5%, 15%, and so on, shown as open boxes.

The region of coincident depressed and elevated that is more prevalent in RI and IN cases supports the combined presence of scattering by frozen hydrometeors along with emission by liquid hydrometeors. Such a presence is indicative of clouds that are not solely warm, which is at odds with the conclusion of KJ12 regarding the importance of warm rain in RI. For a comparison of with the cyan region used by KJ12, each of the TMI cases cited as containing a predominantly cyan ring from the first two sections of KJ12’s Table 2 are displayed in Fig. 10. Rita at 1500 UTC 20 September 2005 is the only case excluded, because of its immediate proximity to Cuba introducing the potential for topographical interaction. In these products, while ≤ 250 K is often associated with pink regions indicative of deep convection following Lee et al. (2002), they are by no means exclusive. Some examples of associated with a lack of nearby deep convection include the southwestern inner core and rainbands extending south and west of Fabian (Fig. 10a), isolated regions south and west of Danielle’s eye in addition to the entire rainband southeast of the center (Fig. 10b), east and north of Frances’ inner core (Fig. 10c), the western half of Karl’s inner core (Fig. 10d), northwest of Dennis’s inner core and farther north within its rainbands (Fig. 10e), and the rainband wrapping east and south of Florence (Fig. 10f). Each of these aforementioned scenarios is unable to be fully accounted for as stratiform precipitation as a result of the extended distances between depressions and pink regions. One region of note, however, includes the rainband wrapping east and south of Florence, with the southernmost portion of this rainband classified as cyan and possessing ≤ 250 K. Two plausible physical explanations for this pattern are either the deep convection to the north’s stratiform region extending south by 100–200 km yielding the cyan region (unlikely because of the vast distance), or a progression from more vertically developed convection with associated ice processes in the north to shallower, developing convection farther south (i.e., congestus and shallow cumuli). Regardless, neither of these hypothetical scenarios supports the exclusive association of cyan with low-level clouds or warm rain. Also of note here is that the ring algorithm in the  appendix objectively indicates predominantly cyan rings in four of the six cases (Figs. 10a,b,e,f) while rings are present in four cases (Figs. 10a,c,d,e). Both the and predominantly cyan ring methods match examples that KJ12 associated with RI (Figs. 10a,b,c,e), while rejecting one of the two delayed RI episodes (i.e., RI had already begun; Figs. 10d,f).

Fig. 10.

TMI 37-GHz regime classification per Fig. 1 overlaid with regions of ≤ 250 K (white outline) of the (a) Fabian, (b) Danielle, (c) Frances, (d) Karl, (e) Dennis, and (f) Florence cases from KJ12 with the date and time for each overpass indicated in the panel titles. False color regime classifications for cyan and pink regions are those colors, respectively, while “other” designations are displayed as dark green. Intensities at each respective overpass are 33, 23, 23, 28, 36, and 26 m s−1.

Fig. 10.

TMI 37-GHz regime classification per Fig. 1 overlaid with regions of ≤ 250 K (white outline) of the (a) Fabian, (b) Danielle, (c) Frances, (d) Karl, (e) Dennis, and (f) Florence cases from KJ12 with the date and time for each overpass indicated in the panel titles. False color regime classifications for cyan and pink regions are those colors, respectively, while “other” designations are displayed as dark green. Intensities at each respective overpass are 33, 23, 23, 28, 36, and 26 m s−1.

5. Precipitative ring relationship to intensity change

Given the apparent discrepancies between the findings of HN11 and KJ12, in addition to the potential influences of subjective ring quantification on the results of KJ12 revealed in Fig. 10, utilizing the objective ring quantification method in the  appendix is desirable for objectively quantifying any link between such features and RI as supported by Edson and Ventham (2008), HN11, and KJ12. Forecast statistics are evaluated globally for precipitative ring association in various PM products relative to intensity change in Table 2. The statistics are evaluated for the PM product indicated in the header, for the association of a ring with population one relative to the lack of a ring in population two. Here, combined cyan and pink (cyan+pink) rings are used following KJ12, with the stipulation that the precipitation portion of the ring must be ≥50% cyan. As stated in section 2b, we note that the cyan+pink ring methodology employed here is not identical to that of KJ12, in that their methods do not necessitate a noncyan or nonpink region at the TC center and instead employ a radial-gradient-like analysis (H. Jiang 2015, personal communication). Probability of detection (POD) is the number of correct event forecasts (with the event being the ring associated with population 1), relative to the total number of occurrences (all population 1 events). POD values are greatest for each of the populations for rings and lowest for cyan+pink rings by approximately 40% for RI versus IN and RI versus non-RI cases. The POD values for the and rings exceed those of KJ12 for cyan+pink rings in the North Atlantic, while the cyan+pink ring globally does not. False alarm ratio (FAR) is the number of times a precipitative ring is seen with population 2, relative to all precipitative ring occurrences. FAR values exceed 0.75 (0.9) for RI versus IN (non-RI) in all PM ring products, with the best performance at and worst for cyan+pink rings. The Heidke skill score (HSS; Wilks 2006) represents the number of events correctly forecast (ring association with population 1 and lack of a ring associated with population 2) minus the expected number of correct forecasts (half the total sample here, given the two-class forecast) divided by the total number of forecasts minus the expected number of correct forecasts. HSS values are best using for RI versus IN, while cyan+pink rings exhibit the greatest HSS for RI versus non-RI and RI and IN versus SS and WE. Because of the large quantity of cases lacking precipitative rings that are associated with population 2, the HSS values are relatively large. Since RI makes up approximately 4% of (Table 1), one way to approximate a persistence baseline would be to assume that population 2 is forecast for all events regardless of ring presence and, then, to take the HSS difference of this scenario relative to the original HSS. This yields , which is indicative of how much skill the ring association adds to the forecast. While all values are positive, the lowest are for the cyan+pink ring for all three of the scenarios, indicative of the cyan+pink ring adding the least information to the forecast. Finally, a chi-squared test is used to evaluate the robustness of association with a ring in population 1 relative to population 2. First, it is apparent the precipitative ring presence in from HN11 is a robust predictor with values significant at ≥99.99% in all three scenarios. Second, HN11 was incorrect in dismissing the 37-GHz ring presence importance, with ring features also robust for the three scenarios at ≥99.99%. The cyan+pink ring presence (Table 2) is robust at 99.99% for TCs undergoing RI versus non-RI and TCs that will undergo intensification to some extent versus SS and WE cases. However, cyan+pink rings are unable to significantly distinguish between TCs undergoing RI and those intensifying but at a lesser rate, with 18% of both RI and IN cases having such rings. This is at odds with the main finding of KJ12, with possible reasons for the discrepancy being KJ12 not requiring a qualitatively different center appearance relative to the ring, the varied dataset date ranges and sensors utilized, KJ12’s restrictions on ring width and azimuthal coverage that are neglected here, or KJ12 not using cyan+pink rings to predict RI occurrence and instead stating that the ring has to appear at some point during RI. One factor not causing this discrepancy is that while KJ12 investigate only the North Atlantic, repeating Table 2 for only this basin negligibly impacts the forecasting statistics and significance levels (not shown).

Table 2.

POD, FAR, HSS, HSS, and chi-squared test statistical significance levels based on association of a precipitative ring with the first populations and lack of a precipitative ring associated with the second population for the PM product listed in the top row.

POD, FAR, HSS,  HSS, and chi-squared test statistical significance levels based on association of a precipitative ring with the first populations and lack of a precipitative ring associated with the second population for the PM product listed in the top row.
POD, FAR, HSS,  HSS, and chi-squared test statistical significance levels based on association of a precipitative ring with the first populations and lack of a precipitative ring associated with the second population for the PM product listed in the top row.

The question still arises as to the hydrometeoric nature of the cloud populations represented by the rings observed as robust in and from Table 2. To investigate, statistics can be isolated exclusively across the ring region (shown in brown in Fig. A1e), opposed to across the entirety of the TC inner core, as done in section 3. Some cases (e.g., the western portion of Fig. A1e), lack clear distinction between the TC inner core and rainbands. In an effort to limit rainband influences, statistics for ring features are evaluated relative to only ≤100-km radius. This does not remove all rainband impacts on the analyses; however, it is a compromise solution between capturing the full ring region for large TCs despite increased rainband influences for smaller TCs. Subsequent analyses in this section are performed at 1-km resolution, as with the ring algorithm in the  appendix; however, a nearest-neighbor interpolation is utilized to preserve the 8-km character of the original HURSAT-MW dataset.

First, bivariate PDFs of and are reproduced for ring regions in Fig. 11. Differences between the intensity changes are generally minor, aside from the RI PDF (Fig. 11a) spanning a reduced range of and relative to the others. Assuming the 624-case sample is sufficient, this perspective supports the lack of extremely vigorous convection for RI storms (associated with depressed and ), suggested by DeMaria et al. (2012). The distribution peaks for rings are associated with of 220–250 K and of 255–270 K, corresponding to the combined presence of frozen and liquid hydrometeors in cold clouds. The comparable analysis for rings is shown in Fig. 12 with small differences between the intensity change classifications aside from RI. Again, peak frequencies occur between of 220–250 K and of 260–270 K, associated with cold clouds and the combined presence of frozen and liquid hydrometeors. Revisiting the 250- and 260-K thresholds for delineating warm and cold clouds, rings see cold (warm) cloud presences of 51% (28%) for RI, 50% (29%) for IN, 45% (35%) for SS, and 42% (38%) for WE. This analysis further links greater TC intensification to an increased presence of frozen hydrometeors, while rings that are intensifying but not undergoing RI possess more warm clouds than those subsequently undergoing RI and the greatest proportion of warm clouds is associated with WE TCs.

Fig. 11.

As in Fig. 6, but for the innermost 100 km of rings.

Fig. 11.

As in Fig. 6, but for the innermost 100 km of rings.

Fig. 12.

As in Fig. 6, but for the innermost 100 km of rings.

Fig. 12.

As in Fig. 6, but for the innermost 100 km of rings.

Next, quantile–quantile plots of pixels accounting for rings by intensity change are taken for the frequency of ≤ 250 K, ≤ 220 K, and cyan coverage in Fig. 13. For rings, ≤ 250-K coverage (Fig. 13a) exhibits median values of near 50% across each intensity change; however, RI and IN cases exhibit greater coverage of ≤ 250 K than SS or WE systems (i.e., they are shifted toward the top left of the one-to-one line and SS and WE distributions). This reveals rings are typically made up of the joint presence of liquid and frozen hydrometeors within the atmospheric column nearly half of the time, while RI and IN cases are associated with greater ice presence than in lesser intensity changes. Differences in the non-RI 250-K coverage here are all statistically significant at ≥99% per a Kolmogorov–Smirnov test. Results for ≤ 220 K within rings (Fig. 13b) show less-distinguishable differences due to the reduced coverage of these s; yet, the RI and IN cases shift once more toward greater coverage of deep convection for the lowest 85% of the distribution relative to the SS or WE cases, although this shift is not robust. Cyan coverage within rings (Fig. 13c) is generally similar among intensity changes; however, a shift occurs near the 45th percentiles (frequencies of approximately 75%) of the distributions where the SS and WE cases begin to experience greater coverage than the RI or IN cases. Differences among the cyan distributions here are robust at ≥99% for RI versus IN distributions and RI versus non-RI.

Fig. 13.

As in Fig. 9, but for the frequency of pixels collocated with rings that are characterized as ≤ (a) 250 and (b) 220 K, and (c) cyan pixels.

Fig. 13.

As in Fig. 9, but for the frequency of pixels collocated with rings that are characterized as ≤ (a) 250 and (b) 220 K, and (c) cyan pixels.

Figure 14 shows quantile–quantile perspectives across rings for individual intensity changes relative to the full distributions for ≥ 255 and 260 K as well as cyan frequencies. The median values of ≥ 255 K (Fig. 14a) exceed 90%, which is supportive of the joint present of frozen and liquid hydrometeors within the atmospheric column for these features. Differences among the distributions here are slight and are only significant at ≥95% for RI versus non-RI distributions. The ring coverage of ≥ 260 K (Fig. 14b) shows more differentiation than does the 255-K threshold, with RI cases having reduced coverage above this threshold at percentiles ≥25%. This is likely due to the depression of seen in regions with substantial scattering at from the presence of large cloud ice that can scatter at 37 GHz (i.e., the positive slope seen as increases along with for the hydrometeor mode apparent in Fig. 5). The distribution of the RI frequency of ≥ 260 K in the rings is statistically significant at ≥99% relative to both the IN and non-RI distributions. Finally, the cyan frequency within the rings (Fig. 14c) has median values ≥50% regardless of intensity change, but the greatest values are found in the RI and IN cases at 58% and 57%, respectively. This implies appreciable precipitation-size ice presence collocated with cyan regions. The RI and IN distributions are generally intermingled throughout the analysis, implying little difference between the two; however, the shift toward the top left of the figure for those data implies greater cyan coverage being linked to general TC intensification, akin to the results of cyan+pink rings inability to robustly distinguish between RI and IN cases from Table 2. Interestingly, above the 85th percentile the distributions converge; however, the full set of RI distribution differences are robust relative to the IN and non-RI distributions within the rings at ≥99%.

Fig. 14.

As in Fig. 13, but for frequency of pixels collocated with rings that are characterized as: ≥ (a) 255 and (b) 260 K, and (c) cyan pixels.

Fig. 14.

As in Fig. 13, but for frequency of pixels collocated with rings that are characterized as: ≥ (a) 255 and (b) 260 K, and (c) cyan pixels.

6. Summary and conclusions

This article serves to investigate the relative prevalence and importance of warm clouds (i.e., shallow cumuli and cumulus congestus) versus cold clouds (i.e., deep convection or stratiform precipitation) within the TC inner core toward intensification using a 22-yr record of PM observations. Much of this work is done against the backdrop of HN11 and KJ12 with their contrasting perspectives supporting the relative importance of cold versus warm clouds, respectively, for RI episodes. The work further serves to complement the radar-based study of Tao and Jiang (2015). While the work here is substantial, it is noted that TRMM has been replaced by the Global Precipitation Measurement platform while the SSM/I platforms are in the process of being revamped as the Special Sensor Microwave Imager/Sounders, with each of the new sensors having differing frequencies than those analyzed here. Hence, subsequent works should strive to extend the work here to currently operational platforms to maximize the utility of these results, while still considering differences in sensor strategies. The main points of this study are as follows:

  • The pervasive characterization noted throughout of depressed is associated with increased ice presence linked to greater TC intensification. This ice is noted in the combined presence of increased , resulting from emission due to rain and cloud liquid water. The elevated values across all but the most extremely depressed values support the lack of a substantial presence of large ice hydrometeors (i.e., graupel and hail). This finding supports the importance of either deep convection with insufficient large ice production to scatter at 37 GHz and stratiform precipitation with TC intensification. These results supporting the combined presence of liquid and frozen hydrometeors within the atmospheric column throughout the TC inner core are also consistent with the median convective eyewall reflectivities of Cecil et al. (2002).

  • As a result of the smaller range of values within the PDFs for precipitative rings in and in RI cases relative to the full distribution, it appears the premise of DeMaria et al. (2012) regarding RI cases rarely possessing extreme convection is valid.

  • The cyan+pink 37-GHz rings, as defined herein, that surround regions that are not cyan or pink are found to be able to objectively discriminate RI from non-RI, but not RI from IN. Conversely, and rings can differentiate between RI and non-RI, RI and IN, and RI or IN versus SS or WE. The differences between and the cyan+pink rings may be a result of the limited range accounted for by the false color product, which makes objective identification in the values difficult. Radiative transfer simulations reveal that sea surface roughness may be ambiguous from hydrometeor signatures or near-surface wind signatures that may be a response to intensification rather than a harbinger. Such wind influences may be a source of the limited radial gradient seen in 37-GHz frequency analyses over the course of RI that can lead to their lack of a distinctive precipitative ring.

  • It appears the primary source of discrepancies between the physical results of HN11 and KJ12 stems from the description of cyan regions used by KJ12. The KJ12 definition for cyan is “low-level water clouds and warm rain,” while instead Lee et al. (2002) describe cyan as representing “low-level water clouds and rain,” with the key addition of the warm descriptor altering the meaning and interpretation of the cyan regions. Cyan regions merely indicate an absence of appreciable scattering by large frozen hydrometeors (i.e., graupel and hail), but this provides incomplete information about the nature of the source (cloud water, “warm” rain, rain with 85-GHz ice scattering that could be convective or stratiform, or nonhydrometeor influences as discussed in the previous point). Further, the coarse resolution at 37 GHz relative to 85 GHz provides a greater opportunity for nonhomogeneous beam filling to confound interpretation. Cyan regions and appreciable ice scattering at 85 GHz are found to not be mutually exclusive, thus supporting the inferences of HN11 and KJ12, while supporting the physical interpretations from HN11 of the inner core of the RI systems “containing ice and mixed phase microphysical processes…” The recent study of Tao and Jiang (2015) appears to further confirm this paradigm via TRMM PR analyses.

  • While Tao and Jiang (2015) argue against the relationship between 85-GHz rings and RI, it is important to note that they do not describe their methods, and the comparisons they make between an 85-GHz ring with ≤ 250 K and a combined cyan and pink ring are inconsistent as a result of the threshold placed at 85 GHz, while instead a gradient is utilized with the false color product. For a clearer comparison, methods such as those used here are encouraged (i.e., requiring a ring rather than a circular region with common false color appearance throughout) or if utilizing a radial gradient method to evaluate 85 GHz in that manner relative to the 37-GHz false color product. Given such a comparison within a common framework, as done here, we emphasize that or rings appear superior in terms of predicting RI relative to the combined cyan+pink rings surrounding an area that is not cyan or pink.

  • The nature of the cyan regions can be further evaluated using combined radar–radiometer observations. While Tao and Jiang (2015) used the TRMM Precipitation Radar for such analyses, that sensor may underestimate vertical cloud development as a result of a minimum detectable reflectivity of 17–18 dBZ (Kummerow et al. 1998) that cannot capture low-reflectivity regions aloft. Instead, the AMSR2 and CloudSat Cloud Profiling Radar (CPR) appear better equipped, because of the CPR’s sensitivity of −30 dBZ to better characterize the vertical structure of TC inner cores (e.g., Schubert and McNoldy 2010). Finally, modeling studies such as Wang (2014), Terwey and Rozoff (2014), and Harnos and Nesbitt (2016) utilizing three-dimensional analyses at the individual updraft scale can directly categorize the role and presence of cloud populations in TC intensification from both dynamical and thermodynamic perspectives, something current remote sensing observations cannot accomplish.

Acknowledgments

The authors thank Ken Knapp for providing the HURSAT-MW database for SSM/I in addition to code enabling the TMI data to be adapted to a common format. TRMM 1B11 data were provided by NASA’s Mirador repository. Discussions with and feedback from Zhuo Wang, Greg McFarquhar, and Ryan Sriver greatly improved the quality of this manuscript. Guidance from Editor Pat Harr, in addition to Haiyan Jiang and two anonymous reviewers, is also appreciated in refining the manuscript. Funding for this work was provided via NASA under Ramesh Kakar through Hurricane Science Research Program Grant NNX09AB82G, Earth and Space Science Fellowship NNX10AP50H, and Precipitation Measurement Missions Grants NNX13AF86G and NNX16AB70G.

APPENDIX

Objective Scene Interpretation

A method is sought to objectively recenter overpasses when sufficient TC structure exists as a result of errors from limited temporal resolution in best track data and nonlinear TC motions. Furthermore, some quantification of ring presence and characterization is desired to further investigate the results of HN11 and KJ12. We seek a novel method to combine computational efficiency with the ability to accurately recenter overpasses and quantify ring presence. Other methods exist to recenter TCs and quantify ringlike structures, such as the Automated Rotational Center Hurricane Eye Retrieval (ARCHER; Wimmers and Velden 2010, 2016).

A ring in HN11 is loosely categorized as a region of s associated with hydrometeors surrounding a region associated with a lack of hydrometeors in the circulation center vicinity, while for KJ12 it relies solely upon looking for a shift in the radial gradient of the false color product (H. Jiang 2015, personal communication). The HN11 approach can be exploited for automated detection through the precipitation feature framework (Nesbitt et al. 2000). Very simply, the algorithm for each overpass determines within each TC inner core whether any contiguous regions associated with an absence of hydrometeors exist entirely within a contiguous region of s associated with hydrometeors. An example is shown in Fig. A1 and is detailed in the following paragraph.

Fig. A1.

TMI (K) ring quantification of Hurricane Paloma at 1149 UTC 8 Nov 2008. See  appendix text for description of the steps within each panel.

Fig. A1.

TMI (K) ring quantification of Hurricane Paloma at 1149 UTC 8 Nov 2008. See  appendix text for description of the steps within each panel.

First, the raw HURSAT-MW overpass is subset to only the innermost 208 km × 208 km (51 pixels at 8-km resolution; Fig. A1a) and bilinearly interpolated onto a 1-km grid (Fig. A1b). Next, regions associated with hydrometeors are isolated to define possible ring boundaries. A minimum contiguous (four sided) area ≥100 km2 is required, with regions associated with precipitation defined as ≥ 260 K, ≤ 250 K (potential objects colored uniquely in Fig. A1c), or 37-GHz false color combined cyan and pink (cyan+pink; with the region having to be ≥50% cyan following KJ12) regions (as defined by the black boundaries in Fig. 1). Then, regions not associated with hydrometeors are defined that are ≥50 km2. The thresholds defined as lacking hydrometeors are ≤ 255 K, ≥ 270 K (possible features colored uniquely in Fig. A1d), or 37-GHz false color regions that are not cyan+pink (i.e., other). Focus then returns to the precipitation features associated with hydrometeors defined in Fig. A1c with convex hulls generated around each (Fig. A1e), acting to loosely define the hydrometeor regions (allowing for small voids to be neglected; thus, a ring need not be 100% complete but rather substantially complete in azimuth). The convex hull functions as a rubber band would if stretched beyond the perimeter of an object before being allowed to return to form around that object; where after, the rubber band may not be entirely in contact with the object but may have slight gaps between the rubber band and object’s surface. Each precipitation feature lacking hydrometeors is evaluated to determine whether it resides completely within any convex hull (Fig. A1f) with the example’s red feature here residing within the largest convex hull. This feature can then have its centroid taken to be the TC center (black × in Fig. A1g), with the overpass flagged as potentially containing a precipitative ring in the corresponding PM product (the 250-K contour of is shown in black in Fig. A1g for comparison). All potential precipitative rings are subsequently manually double-checked to confirm ring validity around the TC center (5052 , 2987 , and 6660 for a combined 37-GHz false color ring of cyan and pink that is ≥50% cyan). Potential miscategorization arises primarily from edge effects of the analysis domain or the presence of a moat associated with secondary eyewalls. Objectively centered overpasses using the centroids of verified precipitative rings from this methodology are used henceforth. Noted once more is that the precipitating region defining the ring (e.g., Fig. A1e) does not have to be 100% complete in azimuth, but merely needs to allow the convex hull to fully enclose the region associated with a lack of hydrometeors (e.g., the red region in Fig. A1f). Any multifrequency analyses for overpasses with a ring detected in both and use the center. It is again noted that the native 37-GHz resolution of the SSM/I (37 km × 28 km) may be limited in resolving the TC inner core consistently, which can impact the use of the recentering algorithm. For example, Table 5 in Vigh et al. (2012) suggests 25% of initial eye developments for Atlantic TCs would not be able to be resolved by the SSM/I.

REFERENCES

REFERENCES
Adler
,
R. F.
, and
E. B.
Rodgers
,
1977
:
Satellite-observed latent heat release in a tropical cyclone
.
Mon. Wea. Rev.
,
105
,
956
963
, doi:.
Alliss
,
R. J.
,
S.
Raman
, and
S.
Chang
,
1992
:
Special Sensor Microwave/Imager (SSMI) observations of Hurricane Hugo (1989)
.
Mon. Wea. Rev.
,
120
,
2723
2737
, doi:.
Barnes
,
C. E.
, and
G. M.
Barnes
,
2014
:
Eye and eyewall traits as determined with the NOAA WP-3D lower-fuselage radar
.
Mon. Wea. Rev.
,
142
,
3393
3417
, doi:.
Bister
,
K.
, and
K. A.
Emanuel
,
1997
:
The genesis of Hurricane Guillermo: TEXMEX analyses and a modeling study
.
Mon. Wea. Rev.
,
125
,
2662
2682
, doi:.
Black
,
M. L.
,
J. F.
Gamache
,
F. D.
Marks
Jr.
,
C. E.
Samsury
, and
H. E.
Willoughby
,
2002
:
Eastern Pacific Hurricanes Jimena of 1991 and Olivia of 1994: The effect of vertical shear on structure and intensity
.
Mon. Wea. Rev.
,
130
,
2291
2312
, doi:.
Black
,
R. A.
,
H. B.
Bluestein
, and
M. L.
Black
,
1994
:
Unusually strong vertical motions in a Caribbean hurricane
.
Mon. Wea. Rev.
,
122
,
2722
2739
, doi:.
Black
,
R. A.
,
R. W.
Burpee
, and
F. D.
Marks
Jr.
,
1996
:
Vertical motion characteristics of tropical cyclones determined with airborne Doppler radial velocities
.
J. Atmos. Sci.
,
53
,
1887
1909
, doi:.
Cecil
,
D. J.
, and
E. J.
Zipser
,
1999
:
Relationships between tropical cyclone intensity and satellite-based indicators of inner-core convection: 85-GHz ice-scattering signature and lightning
.
Mon. Wea. Rev.
,
127
,
103
123
, doi:.
Cecil
,
D. J.
,
E. J.
Zipser
, and
S. W.
Nesbitt
,
2002
:
Reflectivity, ice scattering, and lightning characteristics of hurricane eyeballs and rainbands. Part I: Quantitative description
.
Mon. Wea. Rev.
,
130
,
769
784
, doi:.
Cocks
,
S. B.
,
I.
Johnson
,
R.
Edson
,
M. A.
Lander
, and
C. P.
Guard
,
1999
: Techniques for incorporating SSM/I imagery into Dvorak tropical cyclone intensity estimates. Preprints, 23rd Conf. on Hurricanes and Tropical Meteorology, Dallas, TX, Amer. Meteor. Soc., 8C.7.
Deirmendjian
,
D.
,
1964
:
Scattering and polarization properties of water clouds and hazes in the visible and infrared
.
Appl. Opt.
,
3
,
187
196
, doi:.
DeMaria
,
M.
,
R. T.
DeMaria
,
J. A.
Knaff
, and
D.
Molenar
,
2012
:
Tropical cyclone lightning and rapid intensity change
.
Mon. Wea. Rev.
,
140
,
1828
1842
, doi:.
DeMaria
,
M.
,
C. R.
Sampson
,
J. A.
Knaff
, and
K. D.
Musgrave
,
2014
:
Is tropical cyclone intensity guidance improving?
Bull. Amer. Meteor. Soc.
,
95
,
387
198
, doi:.
Didlake
,
A. C.
, Jr.
, and
R. A.
Houze
Jr.
,
2013
:
Dynamics of the stratiform sector of a tropical cyclone rainband
.
J. Atmos. Sci.
,
70
,
1891
1911
, doi:.
Dunion
,
J. P.
,
2011
:
Rewriting the climatology of the tropical North Atlantic and Caribbean Sea atmosphere
.
J. Climate
,
24
,
893
908
, doi:.
Dvorak
,
V. F.
,
1975
:
Tropical cyclone intensity analysis and forecasting from satellite imagery
.
Mon. Wea. Rev.
,
103
,
420
430
, doi:.
Edson
,
R. T.
,
2004a
: Tropical cyclone analysis techniques from QUIKSCAT NRCS, wind and ambiguity data and microwave imagery. Preprints, 26th Conf. on Hurricanes and Tropical Meteorology, Miami, FL, Amer. Meteor. Soc., P1.8. [Available online at https://ams.confex.com/ams/pdfpapers/75828.pdf.]
Edson
,
R. T.
,
2004b
: Uses of microwave imagery as a supplement to the Dvorak technique, an integrated technique. Preprints, 26th Conf. on Hurricanes and Tropical Meteorology, Miami, FL, Amer. Meteor. Soc., 8A.2. [Available online at https://ams.confex.com/ams/pdfpapers/75823.pdf.]
Edson
,
R. T.
, and
M. A.
Lander
,
2002
: Evaluation of microwave imagery in the life cycle of tropical cyclones. Preprints, 25th Conf. on Hurricanes and Tropical Meteorology, San Diego, CA, Amer. Meteor. Soc., 12A.2. [Available online at https://ams.confex.com/ams/pdfpapers/35316.pdf.]
Edson
,
R. T.
, and
J. D.
Ventham
,
2008
: Signs of rapid intensification as depicted in the microwave imagery. Preprints, 28th Conf. on Hurricanes and Tropical Meteorology, Orlando, FL, Amer. Meteor. Soc., 15A.2. [Available online at https://ams.confex.com/ams/pdfpapers/137258.pdf.]
En-Bo
,
W.
, and
G.
Yong
,
2005
:
A microwave emissivity model of sea surface under wave breaking
.
Chin. Phys.
,
14
,
1259
1264
, doi:.
Gasiewski
,
A. J.
,
1993
: Microwave radiative transfer in hydrometeors. Atmospheric Remote Sensing by Microwave Radiometry, M. A. Janssen, Ed., John Wiley and Sons, 91–144.
Grody
,
N. C.
, and
F.
Weng
,
2008
:
Microwave emission and scattering from deserts: Theory compared with satellite measurements
.
IEEE Trans. Geosci.
,
46
,
361
375
, doi:.
Harnos
,
D. S.
, and
S. W.
Nesbitt
,
2011
:
Convective structure in rapidly intensifying tropical cyclones as depicted by passive microwave measurements
.
Geophys. Res. Lett.
,
38
,
L07805
, doi:.
Harnos
,
D. S.
, and
S. W.
Nesbitt
,
2016
: Varied pathways for simulated tropical cyclone rapid intensification. Part II: Vertical motion and cloud populations. Quart. J. Roy. Meteor. Soc., 142, 1832–1846, doi:.
Hawkins
,
J. D.
,
T. F.
Lee
,
J.
Turk
,
C.
Sampson
,
J.
Kent
, and
K.
Richardson
,
2001
:
Real-time Internet distribution of satellite products for tropical cyclone reconnaissance
.
Bull. Amer. Meteor. Soc.
,
82
,
567
578
, doi:.
Hazelton
,
A. T.
, and
R. E.
Hart
,
2013
:
Hurricane eyewall slope as determined from airborne radar reflectivity data: Composites and case studies
.
Wea. Forecasting
,
28
,
368
386
, doi:.
Hence
,
D. A.
, and
R. A.
Houze
Jr.
,
2011
:
Vertical structure of hurricane eyewalls as seen by the TRMM Precipitation Radar
.
J. Atmos. Sci.
,
68
,
1637
1652
, doi:.
Hendricks
,
E. A.
,
M. T.
Montgomery
, and
C. A.
Davis
,
2004
:
On the role of “vortical” hot towers in formation of Tropical Cyclone Diana (1984)
.
J. Atmos. Sci.
,
61
,
1209
1232
, doi:.
Hendricks
,
E. A.
,
M. S.
Peng
,
B.
Fu
, and
T.
Li
,
2010
:
Quantifying environmental control on tropical cyclone intensity change
.
Mon. Wea. Rev.
,
138
,
3243
3271
, doi:.
Heymsfield
,
G. M.
,
J. B.
Halverson
,
J.
Simpson
,
L.
Tian
, and
T. P.
Bui
,
2001
:
ER-2 Doppler radar investigations of the eyewall of Hurricane Bonnie during the Convection and Moisture Experiment-3
.
J. Appl. Meteor.
,
40
,
1310
1330
, doi:.
Hollinger
,
J.
,
R.
Lo
, and
G.
Poe
,
1987
: Special Sensor Microwave Imager user’s guide. Naval Research Laboratory, 120 pp.
Hoshino
,
S.
, and
T.
Nakazawa
,
2007
:
Estimation of tropical cyclone’s intensity using TRMM/TMI brightness temperature data
.
J. Meteor. Soc. Japan
,
85
,
437
454
, doi:.
Hou
,
A. Y.
, and Coauthors
,
2014
:
The Global Precipitation Measurement Mission
.
Bull. Amer. Meteor. Soc.
,
95
,
701
722
, doi:.
Houze
,
R. A.
, Jr.
,
S. S.
Chen
,
B. M.
Smull
,
W.-C.
Lee
, and
M. M.
Bell
,
2007
:
Hurricane intensity and eyewall replacement
.
Science
,
315
,
1235
1239
, doi:.
Johnson
,
R. H.
,
T. M.
Rickenbach
,
S. A.
Rutledge
,
P. E.
Ciesielski
, and
W. H.
Schubert
,
1999
:
Trimodal characteristics of tropical convection
.
J. Climate
,
12
,
2397
2418
, doi:.
Jones
,
T. A.
,
D. J.
Cecil
, and
M.
DeMaria
,
2006
:
Passive-microwave enhanced Statistical Hurricane Intensity Prediction Scheme
.
Wea. Forecasting
,
21
,
613
635
, doi:.
Kaplan
,
J.
, and
M.
DeMaria
,
2003
:
Large-scale characteristics of rapidly intensifying tropical cyclones in the North Atlantic basin
.
Wea. Forecasting
,
18
,
1093
1108
, doi:.
Kaplan
,
J.
,
M.
DeMaria
, and
J. A.
Knaff
,
2010
:
A revised tropical cyclone rapid intensification index for the Atlantic and eastern North Pacific basins
.
Wea. Forecasting
,
25
,
220
241
, doi:.
Kelley
,
O. A.
,
J.
Stout
, and
J. B.
Halverson
,
2004
:
Tall precipitation cells in tropical cyclone eyeballs are associated with tropical cyclone intensification
.
Geophys. Res. Lett.
,
31
,
L24112
, doi:.
Kieper
,
M.
, and
H.
Jiang
,
2012
:
Predicting tropical cyclone rapid intensification using the 37 GHz ring pattern identified from passive microwave measurements
.
Geophys. Res. Lett.
,
39
,
L13804
, doi:.
Knapp
,
K. R.
, and
J. P.
Kossin
,
2007
:
New global tropical cyclone data set from ISCCP B1 geostationary satellite data
.
J. Appl. Remote Sens.
,
1
,
013505
, doi:.
Knapp
,
K. R.
,
M. C.
Kruk
,
D. H.
Levinson
,
H. J.
Diamond
, and
C. J.
Neumann
,
2010
:
The International Best Track Archive for Climate Stewardship (IBTrACS): Unifying tropical cyclone data
.
Bull. Amer. Meteor. Soc.
,
91
,
363
376
, doi:.
Kummerow
,
C. D.
,
W.
Barnes
,
T.
Kozu
,
J.
Shiue
, and
J.
Simpson
,
1998
:
The Tropical Rainfall Measuring Mission (TRMM) sensor package
.
J. Atmos. Oceanic Technol.
,
15
,
809
817
, doi:.
Lee
,
T. F.
,
F. J.
Turk
,
J.
Hawkins
, and
K.
Richardson
,
2002
:
Interpretation of TRMM TMI images of tropical cyclones
.
Earth Interact.
,
6
, doi:.
Liu
,
G.
,
1998
:
A fast and accurate model for microwave radiance calculations
.
J. Meteor. Soc. Japan
,
76
,
335
343
.
Lonfat
,
M.
,
F. D.
Marks
Jr.
, and
S. S.
Chen
,
2004
:
Precipitation distribution in tropical cyclones using the Tropical Rainfall Measuring Mission TRMM Microwave Imager: A global perspective
.
Mon. Wea. Rev.
,
132
,
1645
1660
, doi:.
McFarquhar
,
G. M.
,
B. F.
Jewett
,
M. S.
Gilmore
,
S. W.
Nesbitt
, and
T.-L.
Hsieh
,
2012
:
Vertical velocity and microphysical distributions related to the rapid intensification in a simulation of Hurricane Dennis (2005)
.
J. Atmos. Sci.
,
69
,
3515
3534
, doi:.
McGaughey
,
G.
,
E. J.
Zipser
,
R. W.
Spencer
, and
R. E.
Hood
,
1996
:
High-resolution passive microwave observations of convective systems over the tropical Pacific Ocean
.
J. Appl. Meteor.
,
35
,
1921
1947
, doi:.
Molinari
,
J.
,
J.
Frank
, and
D.
Vollaro
,
2013
:
Convective bursts, downdraft cooling, and boundary layer recovery in a sheared tropical storm
.
Mon. Wea. Rev.
,
141
,
1048
1060
, doi:.
Nesbitt
,
S. W.
,
E. J.
Zipser
, and
D. J.
Cecil
,
2000
:
A census of precipitation features in the tropics using TRMM: Radar, ice scattering, and lightning observations
.
J. Climate
,
13
,
4087
4106
, doi:.
Nguyen
,
S. V.
,
R. K.
Smith
, and
M. T.
Montgomery
,
2008
:
Tropical cyclone intensification and predictability in three dimensions
.
Quart. J. Roy. Meteor. Soc.
,
134
,
563
582
, doi:.
Rao
,
G. V.
, and
P. D.
MacArthur
,
1994
:
The SSM/I estimated rainfall amounts of tropical cyclones and their potential in predicting the cyclone intensity changes
.
Mon. Wea. Rev.
,
122
,
1568
1574
, doi:.
Rao
,
G. V.
, and
J. H.
McCoy
,
1997
:
SSM/I measured microwave brightness temperature (TB’s), anomalies of (TB’s), and their relationship to typhoon intensification
.
Nat. Hazards
,
15
,
1
19
, doi:.
Rienecker
,
M. M.
, and Coauthors
,
2011
:
MERRA: NASA’s Modern Era Retrospective Reanalysis for Research and Applications
.
J. Climate
,
24
,
3624
3648
, doi:.
Rodgers
,
E. B.
, and
R. F.
Adler
,
1981
:
Tropical cyclone rainfall characteristics as determined from a satellite passive microwave radiometer
.
Mon. Wea. Rev.
,
109
,
506
521
, doi:.
Rodgers
,
E. B.
, and
H. F.
Pierce
,
1995
:
A satellite observational study of precipitation characteristics in western North Pacific tropical cyclones
.
J. Appl. Meteor.
,
34
,
2587
2599
, doi:.
Rodgers
,
E. B.
,
S. W.
Chang
, and
H. F.
Pierce
,
1994
:
A satellite observational and numerical study of precipitation characteristics in western North Atlantic tropical cyclones
.
J. Appl. Meteor.
,
33
,
129
139
, doi:.
Rogers
,
R.
,
2010
:
Convective-scale structure and evolution during a high-resolution simulation of tropical cyclone rapid intensification
.
J. Atmos. Sci.
,
67
,
44
70
, doi:.
Rozoff
,
C. M.
,
W. H.
Schubert
, and
J. P.
Kossin
,
2008
:
Some dynamical aspects of tropical cyclone concentric eyewalls
.
Quart. J. Roy. Meteor. Soc.
,
134
,
583
593
, doi:.
Rutledge
,
S. A.
, and
P.
Hobbs
,
1983
:
The mesoscale and microscale structure and organization of clouds and precipitation in midlatitude cyclones. VIII: A model for the “seeder-feeder” process in warm-frontal rainbands
.
J. Atmos. Sci.
,
40
,
1185
1206
, doi:.
Schluessel
,
P.
, and
H.
Luthardt
,
1991
:
Surface wind speeds over the North Sea from Special Sensor Microwave/Imager observations
.
J. Geophys. Res.
,
96
,
4845
4853
, doi:.
Schubert
,
W. H.
, and
J. J.
Hack
,
1982
:
Inertial stability and tropical cyclone development
.
J. Atmos. Sci.
,
39
,
1687
1697
, doi:.
Schubert
,
W. H.
, and
B. D.
McNoldy
,
2010
:
Application of the concepts of Rossby length and Rossby depth to tropical cyclone dynamics
.
J. Adv. Model. Earth Syst.
,
2
, doi:.
Spencer
,
R. W.
,
H. M.
Goodman
, and
R. E.
Hood
,
1989
:
Precipitation retrieval over land and ocean with the SSM/I. Part I: Identification and characteristics of the scattering signal
.
J. Atmos. Oceanic Technol.
,
6
,
254
273
, doi:.
Stogryn
,
A.
,
1972
:
The emissivity of sea foam at microwave frequencies
.
J. Geophys. Res.
,
77
,
1658
1666
, doi:.
Tao
,
C.
, and
H.
Jiang
,
2015
:
Distributions of shallow to very deep precipitation–convection in rapidly intensifying tropical cyclones
.
J. Climate
,
28
,
8791
8824
, doi:.
Terwey
,
W. D.
, and
C. M.
Rozoff
,
2014
:
Objective convective updraft identification and tracking: Part 1. Structure and thermodynamics of convection in the rain band regions of two hurricane simulations
.
J. Geophys. Res. Atmos.
,
119
,
6470
6496
, doi:.
Velden
,
C. S.
,
T. L.
Olander
, and
R. M.
Zehr
,
1998
:
Development of an objective scheme to estimate tropical cyclone intensity from digital geostationary satellite infrared imagery
.
Wea. Forecasting
,
13
,
172
186
, doi:.
Vigh
,
J. L.
, and
W. H.
Schubert
,
2009
:
Rapid development of the tropical cyclone warm core
.
J. Atmos. Sci.
,
66
,
3335
3350
, doi:.
Vigh
,
J. L.
,
J. A.
Knaff
, and
W. H.
Schubert
,
2012
:
A climatology of hurricane eye formation
.
Mon. Wea. Rev.
,
140
,
1405
1426
, doi:.
Vivekanandan
,
J.
,
J.
Turk
, and
V. N.
Bringi
,
1991
:
Ice water path estimation and characterization using passive microwave radiometry
.
J. Appl. Meteor.
,
30
,
1407
1421
, doi:.
Wang
,
X.
,
Y.
Ma
, and
N. E.
Davidson
,
2013
:
Secondary eyewall formation and eyewall replacement cycles in a simulated hurricane: Effect of the net radial force in the hurricane boundary layer
.
J. Atmos. Sci.
,
70
,
1317
1341
, doi:.
Wang
,
Z.
,
2014
:
Role of cumulus congestus in tropical cyclone formation in a numerical model simulation
.
J. Atmos. Sci.
,
71
,
1681
1700
, doi:.
Wilheit
,
T. T.
,
1986
:
Some comments on passive microwave measurement of rain
.
Bull. Amer. Meteor. Soc.
,
67
,
1226
1232
, doi:.
Wilks
,
D. S.
,
2006
:
Statistical Methods in the Atmospheric Sciences.
Academic Press, 627 pp.
Wimmers
,
A. J.
, and
C. S.
Velden
,
2010
:
Objectively determining the rotational center of tropical cyclones in passive microwave imagery
.
J. Appl. Meteor. Climatol.
,
49
,
2013
2034
, doi:.
Wimmers
,
A. J.
, and
C. S.
Velden
,
2016
:
Advancements in objective multisatellite tropical cyclone center fixing
.
J. Appl. Meteor. Climatol.
,
55
,
197
212
, doi:.
Wood
,
K. M.
, and
E. A.
Ritchie
,
2015
:
A definition for rapid weakening of North Atlantic and eastern North Pacific tropical cyclones
.
Geophys. Res. Lett.
,
42
,
10 091
10 097
, doi:.
Wu
,
R.
, and
J. A.
Weinman
,
1984
:
Microwave radiances from precipitating clouds containing aspherical ice, combined phase, and liquid hydrometeors
.
J. Geophys. Res.
,
89
,
7170
7178
, doi:.
Zhang
,
J. A.
,
R. F.
Rogers
,
P. D.
Reasor
,
E. W.
Uhlhorn
, and
F. D.
Marks
Jr.
,
2013
:
Asymmetric boundary layer structure from dropsonde composites in the presence of environmental wind shear
.
Mon. Wea. Rev.
,
141
,
3968
3984
, doi:.